Urban flood disaster risk assessment method based on crowd sensing, medium and equipment

Abstract

The invention provides an urban flood disaster risk assessment method based on crowd sensing, a medium and equipment, and the method comprises the steps: collecting flood disaster text data on a social platform, carrying out the preprocessing, and constructing a flood disaster text data set; extracting flood disaster common words from the flood disaster text data set, classifying the flood disaster common words according to subject terms, and constructing a disaster descriptive lexicon based on a co-occurrence network; designing a crowdsourcing questionnaire based on the disaster descriptive word bank, issuing the crowdsourcing questionnaire to a target user group, and collecting flood disaster feedback data and geographic information fed back by the target user group; performing spatial distribution on the collected flood disaster feedback data according to geographic information; and carrying out risk assessment by combining two types of indexes which respectively represent the flood disaster situation and the flood disaster influence area. According to the invention, real-time monitoring and accurate assessment of urban flood disasters are realized, and the accuracy and real-time performance of flood disaster risk assessment are improved.

Description

Technical Field

[0001] The present invention belongs to the field of flood disaster information technology, and specifically relates to a method, medium and equipment for urban flood disaster risk assessment based on crowd intelligence perception. Background Art

[0002] Traditional flood disaster risk assessment methods mainly rely on limited monitoring station data, which often have shortcomings such as poor timeliness and limited coverage, making it difficult to fully and real-time grasp the dynamic development of flood disasters.

[0003] In the prior art, the risk assessment of flood disasters usually adopts the following methods: 1. Monitoring station data analysis: Statistical analysis and risk assessment are carried out based on the data from fixed monitoring stations. This method relies on the density of monitoring stations and the timeliness of data transmission, and there are problems of information lag and insufficient spatial coverage. 2. Remote sensing technology: Satellite remote sensing technology is used to obtain a wide range of environmental data for monitoring and assessment of flood disasters. However, remote sensing technology is limited by meteorological conditions and satellite transit time, making it difficult to achieve high-frequency real-time monitoring. 3. Numerical simulation: Based on hydrological models and meteorological forecast data, the occurrence and development of flood disasters are simulated. This method requires a large amount of historical data and precise model parameters, and the accuracy and practicality of the model are subject to certain limitations.

[0004] In recent years, the concept of crowd perception has provided a new solution for the rapid and real-time assessment of urban flood disaster risks. Crowd perception technology can realize real-time monitoring and accurate assessment of flood disasters by collecting widely distributed public perception data, making up for the shortcomings of traditional methods. Relying on the information collection function of smartphones and the interactive communication of online communities, a channel model for uploading user perception data has been formed, which greatly enhances the timeliness and coverage of data. With the development of social media and big data technology, disaster information on social platforms has gradually increased, becoming an important channel for real-time information acquisition and task release. Through text mining technology, key information about flood disasters can be extracted from text data in social media, and a descriptive vocabulary can be formed, which can be further used for flood disaster risk assessment. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention proposes a method, medium and equipment for urban flood disaster risk assessment based on crowd intelligence perception, which can collect and analyze flood disaster related data from the public in real time, and improve the accuracy and timeliness of flood disaster risk assessment.

[0006] To achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for assessing urban flood disaster risk based on crowd intelligence perception, comprising the following steps: Step 1: Collect and preprocess flood disaster text data on social platforms to build a flood disaster text dataset; Step 2: Extract common flood disaster words from the flood disaster text dataset, classify the common flood disaster words according to the subject words, and build a disaster descriptive word library based on the co-occurrence network; Step 3: Design a crowdsourcing questionnaire based on the disaster descriptive vocabulary and publish it to the target user group to collect flood disaster feedback data and geographic information from the target user group; Step 4: Spatially distribute the collected flood disaster feedback data according to geographic information; select indicators representing the flood disaster situation from the flood disaster feedback data, select indicators representing the flood disaster-affected area according to geographic information, and combine these two types of indicators for risk assessment.

[0007] Optionally, step 1 specifically includes the following sub-steps: Step 1.1: Use the words used in the literature to describe urban flood disasters, search and crawl to collect flood disaster text data on social platforms; Step 1.2: Clean the collected flood disaster text data; Step 1.3: Perform word segmentation on the cleaned flood disaster text data; Step 1.4: Delete words irrelevant to flood disasters in the flood disaster text data after word segmentation; Step 1.5: Perform part-of-speech tagging on the words in the flood disaster text data to form a flood disaster text dataset.

[0008] Optionally, step 2 specifically includes the following sub-steps: Step 2.1: Calculate the frequency of occurrence of each word in the flood disaster text dataset, and extract words with a frequency greater than the set value as common flood disaster words; Step 2.2: For the common flood disaster words extracted in step 2.1, use semantic similarity analysis to perform synonym aggregation to generate a flood disaster word set; Step 2.3: Use the LDA topic model to classify the flood disaster word set generated in step 2.2, and calculate the probability distribution of words under each topic category; Step 2.4: First, extract the words in each topic category whose probability is higher than the set value. If the probability of all words in the topic category is lower than the set threshold, the topic category is discarded. Then, select the word with the highest probability in each topic category as the topic word representing the topic category, and the other words as keywords. Step 2.5: Calculate the correlation strength between the subject words and keywords in each subject category, construct a co-occurrence matrix based on the correlation strength, and structure the co-occurrence matrix into a disaster descriptive vocabulary.

[0009] Optionally, step 3 specifically includes the following sub-steps: Step 3.1: Use each subject word in the disaster descriptive word library as the stem content of the multiple-choice question, and use the keywords under each subject word as the options of the multiple-choice question. The options are sorted in descending order according to the correlation strength between the subject word and the keyword; Step 3.2: Design open-ended questions based on the keywords in the disaster descriptive word bank; Step 3.3: Post multiple-choice questions and open-ended questions to the target user group; Step 3.4: Collect flood disaster feedback data from the target user groups in response to the selected questions, and collect the geographic information of the target user groups; Step 3.5: Collect the subjective description text of the target user group's feedback on open-ended questions, and at the same time collect the geographic information of the target user group, extract keywords from the subjective description text and add them to the disaster description vocabulary.

[0010] Optionally, in step 3.5, keywords are extracted from the subjective description text and added to the disaster description word library, specifically: Step 3.5.1: Extract words from the subjective description text, use TF-IDF weighting to calculate the weight of each word and filter out words whose weight is greater than the set value; Step 3.5.2: Calculate the correlation strength between the words selected in step 3.5.1 and the subject words used in the open-ended questions, and retain the words with correlation strength greater than the set value; Step 3.5.3: Add the words retained in step 3.5.2 as keywords into the subject category represented by the subject words in the disaster descriptive vocabulary, and use semantic similarity analysis to perform synonym aggregation.

[0011] Optionally, in step 4, GIS is used to spatially distribute the collected flood disaster feedback data according to geographic information.

[0012] Optionally, in step 4, the indicators characterizing the flood disaster situation include flood depth, flood range and flood time, and the indicators characterizing the flood disaster affected area include regional sensitivity and population density.

[0013] Optionally, in step 4, the risk assessment is to assess the degree of flood risk based on the scores and weights of various indicators in the area to be assessed.

[0014] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the urban flood disaster risk assessment method based on crowd intelligence perception as described in the first aspect.

[0015] In a third aspect, the present invention provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for assessing urban flood disaster risk based on crowd intelligence perception as described in the first aspect is implemented.

[0016] The beneficial effects of the present invention are as follows: the present invention can collect and analyze flood disaster-related data from the public in real time, and by utilizing widely distributed public perception data, it can realize real-time monitoring and accurate assessment of flood disasters, save information collection time and cost, solve the problems of small data collection and inflexible location in traditional methods, and improve the accuracy and real-time performance of flood disaster risk assessment. In the data analysis process, the present invention innovatively introduces the co-occurrence network mechanism to construct a disaster description vocabulary, which can effectively mine relevant words and quantify the relevance. The selected keywords can more comprehensively describe the flood disaster situation, and can be subsequently applied to the design of crowdsourcing questionnaires; at the same time, it also makes full use of the convenience of dynamic update of the disaster description vocabulary based on the co-occurrence network, and can timely add effective keywords extracted from questionnaire feedback into the vocabulary, further enriching the disaster description. In the risk assessment process, the present invention combines flood disaster indicators and flood disaster impact area indicators for risk assessment, considers factors more comprehensively, and improves the assessment effect and visualization. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of the urban flood disaster risk assessment method based on crowd intelligence perception.

[0018] Figure 2 This is the construction flow chart of the LDA topic model.

[0019] Figure 3 It is a schematic diagram of the location for reporting urban flood disaster information.

[0020] Figure 4 It is a visualization distribution map of urban flood disaster inundation information.

[0021] Figure 5 It is an assessed urban flood disaster risk map based on crowd sensing data. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0023] In one embodiment, if Figure 1 As shown, the present invention proposes a method for assessing urban flood disaster risk based on crowd intelligence perception, which mainly includes the following steps: Step 1: Collect and preprocess flood disaster text data on social platforms to build a flood disaster text dataset.

[0024] According to the concept of crowd intelligence perception and the needs of urban flood control, we must first define the descriptive and attribute requirements of the collected data, and build an urban flood corpus based on crowd intelligence perception. In fact, the types of information data that can be used for crowd intelligence perception are diverse, including not only text corpus, but also voice corpus, image corpus, and semantic diversity. However, because the generation cost of text data for information publishers is minimal and the information density is high, in this embodiment, large text information publishing platforms such as Weibo and Xiaohongshu are used as the source of flood disaster text data, and information on urban flooding and major urban flood disaster topics is searched, and it is normalized, and meaningless data is deleted to construct a flood disaster text data set. Table 1 shows the collected original flood disaster text data.

[0025] Table 1 Examples of collected flood disaster text data serial number text 1 With such heavy rain, no matter how good the city's drainage system is, it will cause waterlogging on the road in a short period of time! In the past month, I have encountered such heavy rain on the road several times. It feels like there are a lot of heavy rains this year. Why is that? 2 The rainy season this year was really long and the rain was very heavy. This Monday was no exception. As soon as I woke up, I received the evacuation alert. Then I looked outside the window and saw that there was water again. 3 Because of the rain and water accumulation, I couldn't go to the riverside and was forced to run on the city roads. Then I ran farther and farther without any plan and finally took a taxi home. 4 Recently, Zhongshan City in Guangdong Province has been hit by a fierce rainstorm. The rain poured down continuously, plunging the city into floods. The rainstorm caused serious waterlogging in many towns and streets in Zhongshan City. The rainwater accumulated quickly, turning the roads into rivers, and the water was deep enough to cover people's waists. 5 Many vehicles tried to drive through the water but were eventually forced to stop because the water was too deep. 6 Especially in the past few months, Ganzhou has experienced heavy rains every few days. Now the city is full of water. It is inconvenient to go out for tourism and even to buy things.

[0026] The data on the platform have problems such as different text lengths, different language conciseness, and different combinations of sentence components due to different personal expression habits of users. In order to extract commonalities, some mature analysis methods are to divide long texts into single words, pre-process the text word by word into data characteristics that conform to the field of natural language processing, and compare it with the term library of urban flood disasters to prepare for the subsequent extraction of text themes and condensed evaluation perspectives.

[0027] By deleting redundant blank information and noise in text data and removing stop words defined in the general method, only meaningful content words related to urban flood site information are retained, thereby improving the efficiency of flood information segmentation and semantic understanding.

[0028] The text data selected for collection mainly come from time periods when urban flood disasters are prone to occur. Contents that are too vague in describing the objects and express uncommonly are eliminated. Text data that can reflect the living habits, needs of the masses and the overall risk tolerance of the city are analyzed through text mining methods.

[0029] Step 2: Extract common flood disaster words from the flood disaster text dataset, classify the common flood disaster words according to subject words, and build a disaster descriptive word library based on the co-occurrence network.

[0030] The frequency of occurrence of different words in the flood disaster text dataset is counted, and words with a frequency greater than the set value are extracted as common flood disaster words. This can effectively judge the importance of each word, the user's understanding and usage, and the strength of the description of urban flood disasters.

[0031] For the extracted common flood disaster words, semantic similarity analysis is used to perform synonym aggregation to generate a flood disaster word set. The synonym relationship between words is a fuzzy relationship. Characterizing this relationship is not to simply distinguish whether each group of words has a synonym relationship, but to determine the degree to which the words have a synonym relationship with each other. Through some basic text classification methods such as vector method, the full text can be decomposed into sentence groups containing objects with different meanings, and sentence groups with similar meanings are selected and re-merged into one of the description perspectives under urban flood disasters.

[0032] The generated flood disaster word set is classified using the LDA (Latent Dirichlet Allocation) topic model, and the probability distribution of words under each topic category is calculated. The LDA topic model construction process is as follows: Figure 2 As shown. First, extract the words with probability higher than the set value in each topic category. If the probability of all words in the topic category is lower than the set threshold, the topic category is discarded; then select the word with the highest probability in each topic category as the topic word representing the topic category, and the other words are used as keywords. In this embodiment, the selected topic words mainly include flooding depth, flooding range and flooding time.

[0033] The correlation strength between the subject words and keywords in each subject category is calculated, and a co-occurrence matrix is ​​constructed based on the correlation strength. The co-occurrence matrix is ​​structured into a disaster descriptive word library, and the extracted information is classified according to flood depth, flood range, and flood time. Among them, high-frequency words corresponding to user expressions are set to ensure that the word library format is suitable for subsequent use requirements, such as easy query and integration into other systems. Table 2 shows the subject words of flood depth, flood range, and flood time and the main keywords under them. The keywords are sorted in descending order of relevance.

[0034] Table 2 Descending order of subject word-keyword relevance Flooding range Submergence depth Submerged time intersection Below the upper Hour the way Ankle minute Under the bridge Calf yesterday tunnel knee Several days in a row metro station Waist Depth All night Junction thigh the next day area wheel short Underground Passage Handlebars slow Telephone pole Chassis All night Parking Garage corridor continued square … Too long … …

[0035] Step 3: Design a crowdsourcing questionnaire based on the disaster descriptive vocabulary and publish it to the target user group to collect flood disaster feedback data and geographic information from the target user group.

[0036] The questionnaire should be tailored to the process of urban flood disasters and meet the needs of scenario construction for flood time, location and situation. The questionnaire content should cover information on the initial formation, rapid development and continuous impact of flood disasters, and improve the quality of perceived information and optimize the effect of disaster risk assessment through on-site flooding depth, affected areas and observation time period.

[0037] In terms of questionnaire question design, it is mainly divided into multiple-choice questions and open-ended questions.

[0038] For the selection questions, the subject words in the disaster descriptive word library are used as the stem content of the selection questions, and the keywords under each subject word are used as the options of the selection questions. The options are sorted in descending order according to the correlation strength between the subject words and the keywords. In this way, the structured design of the disaster descriptive word library can be introduced into the crowdsourcing questionnaire design to guide the acquisition of answers with higher relevance to the disaster information. The flood disaster feedback data of the target user group on the selection questions are collected as the data basis for the subsequent flood disaster risk assessment.

[0039] Examples of multiple-choice questions include: What is the current water depth at your location? Please select the closest water depth based on the actual situation.

[0040] A. Below the upper; B. Ankle; C. calf; D. Knee; E. Thigh; F. Waist deep.

[0041] For open-ended questions, open-ended questions are designed based on the subject words in the disaster description word library, and the subjective description texts of the target user group on the open-ended questions are collected. Keywords are extracted from the subjective description texts and added to the disaster description word library. Specifically, the words in the subjective description text are extracted, and the weight of each word is calculated using TF-IDF (Term Frequency-Inverse Document Frequency) weighted to filter out words with a weight greater than the set value; the association strength between the filtered words and the subject words used in the open-ended questions is calculated, and the words with an association strength greater than the set value are retained; the retained words are added as keywords to the subject category represented by the subject words in the disaster description word library, and semantic similarity analysis is used for synonym aggregation. Here, the dynamic update convenience of the disaster description word library based on the co-occurrence network is fully utilized to timely add the effective keywords extracted from the questionnaire feedback to the word library, further enriching the disaster description.

[0042] Examples of open-ended questions include: Please describe the current flooding situation, including the scope, depth and duration of the flooding? For example: A road has been flooded to knee depth, the water level gradually rises, the flooding lasts for several hours, and nearby areas including the bottom of the bridge and underground passages are affected.

[0043] Step 4: Spatially distribute the collected flood disaster feedback data according to geographic information; select indicators representing the flood disaster situation from the flood disaster feedback data, select indicators representing the flood disaster-affected area according to geographic information, and combine these two types of indicators for risk assessment.

[0044] First, the collected flood disaster feedback data is spatially visualized, and visualization schemes are selected as needed. For example, GIS (Geographic Information System) tools can be used to ensure accurate labeling of flood inundation data. GIS technology is used to map the flood-affected areas onto the map, and the inundation depths of different areas are displayed by color or symbols, so as to intuitively present the scope and severity of flood disasters. According to the time slider, the changes in the scope and depth of flood events during their development are displayed, providing users with interactive time series analysis tools to more comprehensively understand the dynamic changes of flood disasters. Figure 3 It is a schematic diagram of the location for reporting urban flood disaster information. Figure 4 This is a visualization distribution map of urban flood disaster inundation information. As can be seen from the figure, through the visualization of spatial distribution, we can preliminarily understand the location and distribution information of flood disasters.

[0045] Then, through a detailed description of the inundation relationship between various objective things in the city and the inundated water bodies, we began to scientifically evaluate the risk level of urban flood disasters and establish an evaluation system for the direct risks caused by urban flood disasters. On the one hand, key indicators for flood risk assessment are extracted from flood disaster feedback data and defined as disaster factor data. On the other hand, indicators that characterize the areas affected by flood disasters are selected based on geographic information, such as combining real-time urban population thermal data. Finally, the weights and eigenvalues ​​of each risk indicator are determined, and an urban flood disaster risk assessment model is constructed to dynamically evaluate the changes in waterlogging risks.

[0046] In this embodiment, the first step is to select appropriate urban flood disaster risk assessment indicators to characterize the flood disaster situation. After using GIS and other methods to clarify environmental information and potential flood risks, determine the risk type and occurrence path, and list and describe the process of risk factors, such as risk sources, probabilities, exposures, consequences, etc. The risk assessment indicators used include flood depth, population density, flood time, and flood range. These indicators have clear quantitative level distinctions and are more objective. These indicators can also cover the flood risk assessment results perceived by the public. This step supplements the scope of the population perceived by the risk assessment system under this step by setting questionnaire options and counting the number of times the options are selected.

[0047] The second step is to grade the degree of individual disaster manifestations and correspond to danger scores. Based on indicators such as flooding time, flooding depth and flooding range, a risk assessment model is constructed in combination with the city's real-time population thermal data. A mathematical model is used to conduct statistical analysis on the sample data to obtain the statistical laws of disaster intensity and loss, thereby realizing risk assessment of natural disasters.

[0048] The third step is to calculate the risk value according to the model, and use the various sensory data and geographic information of regional waterlogging obtained in the above steps to overcome the common problem of insufficient three-dimensional information of submerged water bodies in the coupling method of GIS and remote sensing technology. Based on the natural disaster risk theory, starting from the disaster risk components such as dangerousness and vulnerability, the weights and characteristic values ​​of each indicator are calculated according to the model, and the degree of flood disaster risk in the area is evaluated according to the weights and characteristic values. For example, the flood depth is given a score of Depth: ≥50cm is 10 points, ≥40cm is 7.5 points, ≥27cm is 5 points, and ≥15cm is 2.5 points; the flood time is given a score of Time: ≥3h is 10 points, ≥2h is 7.5 points, ≥1h is 5 points, and ≥0.5h is 2.5 points; the flood range is given a score of Area: underpasses, low-lying areas, subway entrances, underground squares, exhibition halls, schools, civil affairs, etc. are 10 points; ecological / urban construction junction areas, governments, traffic arteries, urban commercial areas, important livelihood municipal facilities, etc. are 7.5 points; general areas are 5 points; areas with more ecology are 2.5 points; determine the population density and give a score of Pop, such as calculating the risk index: Risk = Depth × Time × Area. The weight of each indicator is obtained by analyzing the statistical distribution law of the indicator data. This method overcomes the influence of subjective factors to a certain extent.

[0049] The fourth step is to classify the overall flood risk level. Urban flood risk zoning includes comparing the risk analysis results with pre-set risk criteria or thresholds, or comparing the risk analysis results with each other. Based on the model results, the risks are divided into multiple levels, such as low, medium, and high, and the risk levels are distinguished by colors, with blue representing low risk, yellow representing medium risk, and red representing high risk. Figure 5This is the urban flood disaster risk map obtained through assessment based on crowd intelligence perception data. In the future, based on the urban flood disaster risk assessment results of this step, prevention and diversion suggestions for urban floods before, during and after the flood can be put forward. In addition to being used in the assessment and early warning stage, the collected urban flooding information can also be used as an analytical guidance basis for resource utilization and risk avoidance during emergency management.

[0050] In another embodiment, the present invention proposes a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the urban flood disaster risk assessment method based on crowd intelligence perception of the aforementioned embodiment.

[0051] In another embodiment, the present invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the urban flood disaster risk assessment method based on crowd intelligence perception of the aforementioned embodiment is implemented.

[0052] In the embodiments disclosed in the present application, the computer storage medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. More specific examples of computer storage media may include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0053] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0054] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for assessing urban flood disaster risk based on crowd intelligence perception, characterized in that: The steps include: Step 1: Collect and preprocess flood disaster text data on social platforms to build a flood disaster text dataset; Step 2: Extract common flood disaster words from the flood disaster text dataset, classify the common flood disaster words according to the subject words, and build a disaster descriptive word library based on the co-occurrence network; Step 3: Design a crowdsourcing questionnaire based on the disaster descriptive vocabulary and publish it to the target user group to collect flood disaster feedback data and geographic information from the target user group; Step 4: Spatially distribute the collected flood disaster feedback data according to geographic information; select indicators representing the flood disaster situation from the flood disaster feedback data, select indicators representing the flood disaster-affected area according to geographic information, and combine these two types of indicators for risk assessment.

2. The urban flood disaster risk assessment method based on crowd intelligence perception as claimed in claim 1, characterized in that: The step 1 specifically includes the following sub-steps: Step 1.1: Use the words used in the literature to describe urban flood disasters, search and crawl to collect flood disaster text data on social platforms; Step 1.2: Clean the collected flood disaster text data; Step 1.3: Perform word segmentation on the cleaned flood disaster text data; Step 1.4: Delete words irrelevant to flood disasters in the flood disaster text data after word segmentation; Step 1.5: Perform part-of-speech tagging on the words in the flood disaster text data to form a flood disaster text dataset.

3. The urban flood disaster risk assessment method based on crowd intelligence perception as claimed in claim 1, characterized in that: The step 2 specifically includes the following sub-steps: Step 2.1: Calculate the frequency of occurrence of each word in the flood disaster text dataset, and extract words with a frequency greater than the set value as common flood disaster words; Step 2.2: For the common flood disaster words extracted in step 2.1, use semantic similarity analysis to perform synonym aggregation to generate a flood disaster word set; Step 2.3: Use the LDA topic model to classify the flood disaster word set generated in step 2.2, and calculate the probability distribution of words under each topic category; Step 2.4: First, extract the words in each topic category whose probability is higher than the set value. If the probability of all words in the topic category is lower than the set threshold, the topic category is discarded. Then, select the word with the highest probability in each topic category as the topic word representing the topic category, and the other words as keywords. Step 2.5: Calculate the correlation strength between the subject words and keywords in each subject category, construct a co-occurrence matrix based on the correlation strength, and structure the co-occurrence matrix into a disaster descriptive vocabulary.

4. The urban flood disaster risk assessment method based on crowd intelligence perception as claimed in claim 3 is characterized by: The step 3 specifically includes the following sub-steps: Step 3.1: Use each subject word in the disaster descriptive word library as the stem content of the multiple-choice question, and use the keywords under each subject word as the options of the multiple-choice question. The options are sorted in descending order according to the correlation strength between the subject word and the keyword; Step 3.2: Design open-ended questions based on the keywords in the disaster descriptive word bank; Step 3.3: Post multiple-choice questions and open-ended questions to the target user group; Step 3.4: Collect flood disaster feedback data from the target user groups in response to the selected questions, and collect the geographic information of the target user groups; Step 3.5: Collect the subjective description text of the target user group's feedback on open-ended questions, and at the same time collect the geographic information of the target user group, extract keywords from the subjective description text and add them to the disaster description vocabulary.

5. The urban flood disaster risk assessment method based on crowd intelligence perception as claimed in claim 4 is characterized by: In step 3.5, keywords are extracted from the subjective description text and added to the disaster description word library, specifically: Step 3.5.1: Extract words from the subjective description text, use TF-IDF weighting to calculate the weight of each word and filter out words whose weight is greater than the set value; Step 3.5.2: Calculate the correlation strength between the words selected in step 3.5.1 and the subject words used in the open-ended questions, and retain the words with correlation strength greater than the set value; Step 3.5.3: Add the words retained in step 3.5.2 as keywords into the subject category represented by the subject words in the disaster descriptive vocabulary, and use semantic similarity analysis to perform synonym aggregation.

6. The urban flood disaster risk assessment method based on crowd intelligence perception as claimed in claim 1, characterized in that: In step 4, GIS is used to spatially distribute the collected flood disaster feedback data according to geographic information.

7. The urban flood disaster risk assessment method based on crowd intelligence perception as claimed in claim 1, characterized in that: In step 4, the indicators characterizing the flood disaster situation include flood depth, flood range and flood time, and the indicators characterizing the flood disaster affected area include regional sensitivity and population density.

8. The method for assessing urban flood disaster risk based on crowd intelligence perception as claimed in claim 1, characterized in that: In step 4, the risk assessment is to assess the flood risk level based on the scores and weights of various indicators in the area to be assessed.

9. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables the computer to execute the urban flood disaster risk assessment method based on crowd intelligence perception as described in any one of claims 1 to 8.

10. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for assessing urban flood disaster risk based on crowd intelligence perception as described in any one of claims 1 to 8 is implemented.

Images